Why AI Keeps Guessing About You

You ask for something. The answer arrives in seconds and it is fine — competent, well organised, nothing wrong with it.

Then you spend twenty minutes making it sound like you.

The hours arrive, and then they leave again

This is not a private failing. It is now the most reliably measured thing about AI at work.

In research published by Foxit, executives estimated AI saved them 4.6 hours a week — and reported spending 4 hours and 20 minutes validating what it gave them. Net gain: sixteen minutes. The people below them did worse, landing fourteen minutes a week behind where they started.

Workday found the same shape from another angle: roughly 37% of the time AI saves is given back to correcting and rewriting what it produced.

Almost everyone in these studies reports feeling more productive. The stopwatch disagrees.

The tool is not broken

The natural conclusion is that the models are not good enough yet. It is the wrong one.

Ask for a welcome email for new clients and you will get a welcome email for new clients. It opens warmly, says you are glad they are here, gestures at next steps, closes politely. It is a perfectly good welcome email for the average of everyone who has ever needed one.

That average is what you asked for — not explicitly, but by omission. You did not say who your clients are, what worried them before they signed, what you promised, or how you sound. So the model did the only thing available and answered for the median business.

A generic answer is not a failure. It is the correct answer to a question with nobody in it.

What the rewriting actually is

Watch yourself do the twenty minutes and notice what you are doing. You are not fixing errors. You are supplying facts.

You cut the line about being excited, because you do not talk like that. You add the two-week onboarding, because that is how your business works. You swap "clients" for "members". You delete a promise you cannot keep.

None of that is correction. It is context — arriving late, typed by hand, and thrown away the moment you close the window. Tomorrow you will supply it again.

The cost that does not appear in the hours

The hours are the measurable half, and the smaller one.

The other cost lands when you skip the rewriting — when it is late, or the task is small, or the draft is good enough. That version goes out: competent, unobjectionable, and indistinguishable from what your competitor sent, because it came from the same average.

Do that often enough and the thing that made you worth choosing quietly thins out. Not in any one message. Across all of them.

The question worth asking

The useful question is not "how do I write better prompts". You can write an excellent prompt and still be answering for the median business, because a prompt describes the task — and the gap is in the briefing.

The question is: what would AI need to know about you, once, so that it stops guessing?

That has an answer. It is shorter than you expect, it is written down rather than retyped, and it divides cleanly into two kinds — which is where we go next.